Make your users return before day 1

Summary:

One of the best – and perhaps most overlooked – ways to improve day 1 retention is to design meaningful and compelling callbacks to get your users to return on install day (day 0). You should have at least one callback 2-4 hours after install and having another callback 11-12 hours after install is also a good idea. In order to achieve that with the greatest success possible, you should design an end to your first session. After all, your first session must end at some point in order for users to come back play a second one. Usually, the ideal duration for your first session is going to be between 12 and 18 minutes long. My post this week goes into 3 reasons that justify this

  1. Day 1 is already long-term. The vast majority (usually between 80 and 90%) of installs retained day 1 have returned to the game at least once day 0. If this is something you haven’t looked at before I strongly suggest you do. Trying to get users to come back to play a second session day 0 is a tangible, intermediate, and designable outcome that can in turn help you improve your retention day 1 (and beyond).
  2. In order to have users return to the app day 0, they must have left it at some point. If you look at the percentage of users who return to your game for a second session from the perspective of the duration of the first session, then you see the longer the first session the higher the return rate. However, usually, the return rate plateaus after 18-20 minutes. In other words, users who play 35 minutes on their first session are not much more likely to return than users who play 18 minutes in their first session. Designing an end to your first session can help you set up a meaningful and relevant appointment mechanism that will succeed in making your users come back for a second session on install day.
  3. Very consistently, users return to the app for their second session within hours. It’s safe to bet that 50% of users returning to play a second session in your game will have done so within 4 hours of install. Now, if users return for a second session within a similar time frame – regardless of the genre or specifics of the game – then this probably reflects a constant grounded in the way users interact with mobile entertainment. With that in mind, you will be more successful trying to cater to users’ natural tendencies – rather than counting on your users to change their schedule to accomodate your game. Designing the first callback between 2 and 4 hours of install will increase your chances at being successful at getting users to come back to your mobile game because it matches a natural tendency of your userbase.

 

What makes day 1 retention meaningful (and important)?

Day 1 retention is a (rightfully so) standard metric most mobile game publishers look at to evaluate a title’s early performance. Regardless of the nuances of its definition (is the user returning the next calendar day or between 24 and 48 hours of install?) – day 1 retention is a valuable metric for 2 key reasons. First, it provides a common standard to compare and evaluate a title’s performance. Either a) the evolution of a given title’s performance over time or b) a title’s performance compared to other titles. Second, it does provide meaningful insights in of itself. Day 1 retention is not a purely abstract and artificial metric used to provide a standard of comparison. Having a user return to a game the following day is a strong indication s/he is starting to somehow integrate the game into his/her day-to-day routine.

Improving any metric (positively) is something that requires careful analysis and a well-thought-out course of action. In my experience, retention – and especially day 1 retention – is the hardest KPI to improve. Users will return to a game they enjoy and are drawn to – and that’s the complex (and somewhat intangible) result of the intersection of thematic, game systems, art style and so many other things. It’s not the result of one single feature that can be tuned (excluding any obvious technical or balancing issues), but the sum of all features that makes up a game and defines it.

There is another reason why it can be hard to make in-game changes that improves day 1 retention. When you are considering your title’s performance from the perspective of day 1, day 3 or day 7 timeframes, you are not evaluating your product from a game-centric perspective. Metrics such as tutorial completion, missions initiated, gems spent, stamina used etc. are based on in-game actions. They make sense from the perspective of the game itself and refer to nothing outside of its systems. On the other hand, day 1 is not immediately and meaningfully grounded in the game itself: it simply denotes the passing of time. So, looking at day 1 retention won’t tell you what in-game lever to play with to improve it.

For this post, I would argue that session initiation is one of the most relevant game metric that should be kept in mind while designing the user’s first-time experience and aiming for the highest day 1 retention possible.

Day 1 retention is a meaningful metric to see how the user is integrating the game into his/her day to day, but its horizon is not grounded in the game itself. Day 1 will happen in the user’s life regardless of what’s happening in the game – or whether s/he actually launches your app. Ironically, that’s precisely why day 1 retention matters the most. Any mobile game maker is trying to create a product the user will go back to during his or her busy day; a product that is competing with many others for the user’s (usually limited and fixed) free time. But it’s also because of that very reason that retention performance is probably among the hardest to improve by tweaking game features.

 

Designing for a second session

The focus on improving day 1 retention is widespread. Even a game oriented towards a hardcore audience – which might be able to accommodate itself with lower day 1 retention – will presumably find a big ROI in improving day 1 retention a few percent points. In my experience, the usual levers used to try to increase day 1 retention – the “first reflex” I’ve seen (and displayed) – are tutorials (with the aim of increasing tutorial completion and addressing problematic tutorial steps) and implementing some kind of appointment mechanism for the next day – usually in the form of a daily login bonus. And rightfully so. Improving the tutorial funnel by addressing high drop-off points is bound to have a high ROI for your efforts. Well thought out appointment mechanics that provide a callback the following day can have a significant impact on day 1 retention.

One of the main challenges consists in finding the relevant game/feature metric that correlates with day 1 retention. Doing that allows you to play with the levers you have the most control over to impact the user’s return rate. Very consistently, it appears that day 1 retention is associated to a lower level, intermediate process: namely coming back for a second session day 0. For starters, there are more installs returning to play a second session day 0 than there are installs returning day 1. Depending on the genre, you can see up to 20 percent points more. It’s even more revealing to you shift perspective and look backwards: look at users retained day 1 and their “provenance” (have they returned day 0 or not). You should see that over 80% of users who are retained day 1 have returned to play a second session on day 0.

day_0_and_day_1_return

users_retained_d1_provenance

 

When I came across this I had an important realization. Day 1 is not the earliest (or even the most relevant) timeframe you should be thinking at when designing the user’s first experience. A significant portion of users play more than one session day 0 – even users who eventually don’t end up retained. So, day 1 retention is already “long-term” in some respects. Of course, some users who are retained day 1 will not have returned to play a second session on install day. Conversely, a non-negligible proportion of users who have returned to play a second session on install day will end up not retaining day 1. However, 1) usually less than 20% of users retained day 1 have not returned to play a second session day 0 and 2) day 1 retention is always going to be levels of magnitude higher for users who have returned at least once to the app on day 0. To be clear, this is not a theoretical advice on how to cause your users to retain. It’s the practical observation that you are setting yourself up for success more if you are able to reproduce the conditions the most associated to users retaining. In other words, just aim to recreate the situations most conducive to the desired outcome and let “nature” take over (I’m a big believer of pragmatically embracing the black box and leaving causal conversations for another post).

day1_retention_rate_returned_or_not

 

So why focus on a second session day 0 if the goal is to improve day 1 retention? More often than not, day 1 is the shortest-term timeframe considered when looking at user return patterns. That’s why much too often the day 1 callback many games rely on is a simple login bonus – which is not grounded in any game feature or system, but just in the passing of time. This is not to say you shouldn’t have a daily login bonus (because you definitely should!). Rather, you should not only rely on that as your most linear lever to improve day 1 retention.

Breaking down the day 1 retention phenomenon into simpler parts (namely, returning for a second session day 0) can improve your chances of success. If you define the goal of your first session as leading to another session within 2-4 hours (I would argue conversion should also be another key goal of your first session – but that is something for another topic), then you are providing clear, objective and measurable goals you can design for. Also, you should expect a large number of your installs will play more than 1 session day 0 – more than the number of installs who will retain day 1. That means there is likely to be less resistance to get your users to come back to the game for a second session. By focusing on getting users to return day 0 – as a proxy to increase day 1 retention – you are focusing on something that is both easier to design and likely to reach a higher percent of your installs.

This does not mean that making more users return for a second session day 0 will not cause them to retain better (again, let’s stay pragmatic and leave that in its black box). But having more users returning for a second session means more users are in a situation that is conducive to the desired behavior – namely being retained day 1. For a segment of your userbase, this might result in making some flakier users stick around for one more session – one more day – but ultimately this won’t change their longer-term retention (let alone make them spend in game). However, the more users you are successful in making stick around a bit more than they otherwise would have, the more you are creating an environment conducive to higher day 1 retention.

 

Using this to craft your first user experience

It might sound paradoxical to try to design for users to stop playing at a given point in time. Ultimately, we’re in the business of getting users to play and engage with our games as much as possible. However, the data does indicate that if having users return a second time day 0 correlates positively with day 1 retention – and I most definitely suggest you confirm this with your own title. So designing a – soft – end to your user’s first session can have bigger benefits longer-term. And in this case day 1 retention is not longer-term; it’s just a proxy for that elusive “long-term”. That’s why day 1 retention is important. It’s not the end of the line, it’s at least an indication the user is on-board.

That means that carefully crafting the first session – when it ends and when the second session should begin a few hours after installs – can have the biggest impact on day 1 retention. In other words, carefully designing an end point for the first session in order to create a strong pull to come back and play a second session should be an important – and measurable – goal for you when designing your first session. In order for the callback to be the most meaningful possible, you want to callback for the second session to tie in with the feature that led to your first session to end. One very straightforward (and readable) way to have the first session end is to put some kind of wait timer in front of the user – whether s/he runs out of stamina, or there is a building the user has to wait for, a reward timer the user can look forward to or a game feature that will become available – and determine when the friction point should be removed. The general idea is to determine a) how long your first session should be and b) when is a good time for users to come back for their second session. Some general patterns emerge across midcore games – and if you have a live game, you should be able to determine what might work best for you by looking at how your current users are behaving.

 

When should the first session end?

You can start by looking at your return rate relative to the duration of your first session. Ultimately go with the value that matches the behavior patterns in your game the most. The pattern you are most likely to come across is that the return rate increases a lot with the first few minutes. The increase in return rate gradually slows down the longer the first session.

Let’s look at the example below, the day 1 retention rate of users who play 3 minutes on their first session is 32%. It jumps to 50% for users who play 10 minutes on their first session. So, day 1 retention increased by 18 percent points for an additional 7 minutes. The benefits of having a longer first session are clearly not linear. The D1 retention rate of users who play 17 minutes is not 68% – in the example below it’s 61%. And the trend continues to decrease. The D1 retention for users who play 24 minutes is 67%, etc.

Usually the improvement in return rate decreases the longer the duration of the first session. You want to focus on the early minutes of your first session because the ROI decreases later on. The amount of effort it will take you to get someone to play 30 minutes instead of 25 might not warrant the small increase in return rate you will see from those added 5 minutes. There clearly is a bigger bang for your dev buck to focus on what happens in the first 15-20 minutes of your first session. That doesn’t mean that it’s bad to have your installs play 30 minutes on their first session. Although there can be too long first sessions – where the return rate actually starts decreasing – you need to ensure you have enough users playing that long for those numbers actually mean something. Also, if you’re in a situation where a large portion of your installs are playing 60+ minutes on their first session, then you probably don’t have that big of a problem to begin with.

text_body_return_per_length

 

 

When should the callback for the second session be?

In an ideal world, you want to capitalize on the first session end. If you created a compelling enough reason for users to stop playing their first session, then you should capitalize on that to get them to return to the game for a second session. If you want users to end their first session after 14 minutes, then you can place a friction at that moment – for example, a building users have to wait for, or a chest that will provide them what they need to proceed to the next mission. The next important aspect to determine is how long the building time should be. Of course, you need to have a push notification of some sort to ensure your callback is effective. If your users are looking at their watch and don’t need a notification to know when to return to use the building they’ve been waiting for, then you probably don’t need to invest much effort into getting them back into the game in the first place.

There are multiple factors you can consider to determine when to set the callback. My preferred approach is to see when users are predisposed to come back to the game. Generally, users initiate their second session day 0 very soon. You can expect a majority of users who will return for a second session to do so within 4 hours of install. It’s not uncommon to see a big chunk of users returning by hour 1 or 2 after install. Usually close to 80% of installs who will return to the app will have done so within 24 hours of install.

users_returing_for_a_second_session

 

So that is precisely the kind of data you want to base your design on. If regardless of the game, genre, day of the week, etc. users are consistently returning within a short time frame, that indicates it’s the moment you should be aiming at. At the very least, from a material point of view, that means that there is a good availability window at that moment after an install. From a more behaviorist point of view, this suggests there is a good predisposition or inclination for users to return within that time frame. Again, by setting an appointment 2-4 hours after install, you won’t magically cause your users to return to play a second session. But you are making that callback occur at a heightened moment of availability and are making it easier for users (read: it’s doesn’t require them to act contrarily to their predispositions) to give your game another short and come back.

Another thing you can consider is that a non-negligible portion of your installs will play more than 2 sessions on day 0. You can look at the distribution of your installs per sessions played day 0. Chances are a majority of your returning users are playing more than 2 sessions. The more users come back, the more likely they are to be engaged and like the game, day 1 retention improves.

sessions_played_d0_and_d1_ret

So thinking beyond the second session day 0 can be interesting for a number of reasons. The most important one being that you can leverage your first session to contain more than one callback moment. For example, you can have your users receive a chest early on in the tutorial that will initiate a 12-hour chest (callback 2), and later try to make the first session end with a 2-hour building timer (callback 1). Doing that allows you to a) have multiple chances to get your users to come back play a second session day 0 – if they don’t come back for the building, then maybe they’ll come back to open then chest; and b) incentivize users to come back multiple times day 0. You can see what timing is best for you by looking at what time session 3 occurs. The later/more sessions we are talking, the weaker the pattern is. However, whether it’s for the second or third session, around 12 hours after install seems like an interesting moment to target.

session2_and_3

 

You can always hand craft your first-time experience to exceed 3 sessions on day 0 – but that might be overkill. The main takeaway here is there are huge benefits in getting users to return to your game on install day. Evience shows that should be one of your design goals. And the advantage is, if one of the goals of your first session is to lead to a second session within hours of install, then you are setting yourself clear and measurable goals you can iterate on.

3 comments

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s